adithya8
commited on
Commit
·
8b24411
1
Parent(s):
c9b2661
model upload
Browse files- .gitattributes +0 -0
- README.md +0 -0
- added_tokens.json +1 -0
- config.json +36 -0
- merges.txt +0 -0
- pytorch_model.bin +3 -0
- special_tokens_map.json +1 -0
- tokenizer_config.json +1 -0
- trainer_state.json +1498 -0
- training_args.bin +3 -0
- vocab.json +0 -0
.gitattributes
CHANGED
|
File without changes
|
README.md
CHANGED
|
File without changes
|
added_tokens.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"<|insep|>": 50257}
|
config.json
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "HaRT/outputs/hart_pt_dgx/checkpoint-120000",
|
| 3 |
+
"activation_function": "gelu_new",
|
| 4 |
+
"add_history": true,
|
| 5 |
+
"architectures": [
|
| 6 |
+
"HaRTPreTrainedModel"
|
| 7 |
+
],
|
| 8 |
+
"attn_pdrop": 0.1,
|
| 9 |
+
"bos_token_id": 50256,
|
| 10 |
+
"embd_pdrop": 0.1,
|
| 11 |
+
"eos_token_id": 50256,
|
| 12 |
+
"extract_layer": 11,
|
| 13 |
+
"freeze_model": false,
|
| 14 |
+
"gradient_checkpointing": false,
|
| 15 |
+
"initializer_range": 0.02,
|
| 16 |
+
"layer_ins": 2,
|
| 17 |
+
"layer_norm_epsilon": 1e-05,
|
| 18 |
+
"model_type": "gpt2",
|
| 19 |
+
"n_ctx": 1024,
|
| 20 |
+
"n_embd": 768,
|
| 21 |
+
"n_head": 12,
|
| 22 |
+
"n_inner": null,
|
| 23 |
+
"n_layer": 12,
|
| 24 |
+
"n_positions": 1024,
|
| 25 |
+
"resid_pdrop": 0.1,
|
| 26 |
+
"summary_activation": null,
|
| 27 |
+
"summary_first_dropout": 0.1,
|
| 28 |
+
"summary_proj_to_labels": true,
|
| 29 |
+
"summary_type": "cls_index",
|
| 30 |
+
"summary_use_proj": true,
|
| 31 |
+
"transformers_version": "4.5.1",
|
| 32 |
+
"use_cache": true,
|
| 33 |
+
"use_hart_no_hist": false,
|
| 34 |
+
"use_qh05_wts": false,
|
| 35 |
+
"vocab_size": 50258
|
| 36 |
+
}
|
merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0f518d49be03d0d7f8f3705bebddfa8ae247d69e59e5fce857d53bb0073458f0
|
| 3 |
+
size 628511425
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"bos_token": "<|endoftext|>", "eos_token": "<|endoftext|>", "unk_token": "<|endoftext|>", "sep_token": "<|insep|>"}
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"unk_token": "<|endoftext|>", "bos_token": "<|endoftext|>", "eos_token": "<|endoftext|>", "add_prefix_space": false, "model_max_length": 1024, "special_tokens_map_file": null, "name_or_path": "gpt2"}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1498 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 3.197413921356201,
|
| 3 |
+
"best_model_checkpoint": "HaRT/outputs/hart_pt_dgx/checkpoint-120000",
|
| 4 |
+
"epoch": 5.0,
|
| 5 |
+
"global_step": 120000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.02,
|
| 12 |
+
"learning_rate": 0.00024345226236201876,
|
| 13 |
+
"loss": 4.3078,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.04,
|
| 18 |
+
"learning_rate": 0.00024243363364920699,
|
| 19 |
+
"loss": 3.8284,
|
| 20 |
+
"step": 1000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.06,
|
| 24 |
+
"learning_rate": 0.00024141500493639518,
|
| 25 |
+
"loss": 3.772,
|
| 26 |
+
"step": 1500
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.08,
|
| 30 |
+
"learning_rate": 0.00024039637622358338,
|
| 31 |
+
"loss": 3.7335,
|
| 32 |
+
"step": 2000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.1,
|
| 36 |
+
"learning_rate": 0.00023937774751077158,
|
| 37 |
+
"loss": 3.6854,
|
| 38 |
+
"step": 2500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.12,
|
| 42 |
+
"learning_rate": 0.0002383591187979598,
|
| 43 |
+
"loss": 3.6769,
|
| 44 |
+
"step": 3000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.15,
|
| 48 |
+
"learning_rate": 0.000237340490085148,
|
| 49 |
+
"loss": 3.6263,
|
| 50 |
+
"step": 3500
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.17,
|
| 54 |
+
"learning_rate": 0.0002363218613723362,
|
| 55 |
+
"loss": 3.6499,
|
| 56 |
+
"step": 4000
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.19,
|
| 60 |
+
"learning_rate": 0.00023530323265952442,
|
| 61 |
+
"loss": 3.6191,
|
| 62 |
+
"step": 4500
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.21,
|
| 66 |
+
"learning_rate": 0.00023428460394671262,
|
| 67 |
+
"loss": 3.6148,
|
| 68 |
+
"step": 5000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.23,
|
| 72 |
+
"learning_rate": 0.00023326597523390085,
|
| 73 |
+
"loss": 3.6111,
|
| 74 |
+
"step": 5500
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.25,
|
| 78 |
+
"learning_rate": 0.00023224734652108902,
|
| 79 |
+
"loss": 3.581,
|
| 80 |
+
"step": 6000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.27,
|
| 84 |
+
"learning_rate": 0.00023122871780827724,
|
| 85 |
+
"loss": 3.5799,
|
| 86 |
+
"step": 6500
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.29,
|
| 90 |
+
"learning_rate": 0.00023021008909546544,
|
| 91 |
+
"loss": 3.5888,
|
| 92 |
+
"step": 7000
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.31,
|
| 96 |
+
"learning_rate": 0.00022919146038265367,
|
| 97 |
+
"loss": 3.5619,
|
| 98 |
+
"step": 7500
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.33,
|
| 102 |
+
"learning_rate": 0.00022817283166984186,
|
| 103 |
+
"loss": 3.5589,
|
| 104 |
+
"step": 8000
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.35,
|
| 108 |
+
"learning_rate": 0.00022715420295703006,
|
| 109 |
+
"loss": 3.5657,
|
| 110 |
+
"step": 8500
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.38,
|
| 114 |
+
"learning_rate": 0.0002261355742442183,
|
| 115 |
+
"loss": 3.5589,
|
| 116 |
+
"step": 9000
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.4,
|
| 120 |
+
"learning_rate": 0.00022511694553140646,
|
| 121 |
+
"loss": 3.5379,
|
| 122 |
+
"step": 9500
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.42,
|
| 126 |
+
"learning_rate": 0.00022409831681859468,
|
| 127 |
+
"loss": 3.5361,
|
| 128 |
+
"step": 10000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.44,
|
| 132 |
+
"learning_rate": 0.00022307968810578288,
|
| 133 |
+
"loss": 3.5392,
|
| 134 |
+
"step": 10500
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.46,
|
| 138 |
+
"learning_rate": 0.0002220610593929711,
|
| 139 |
+
"loss": 3.5178,
|
| 140 |
+
"step": 11000
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.48,
|
| 144 |
+
"learning_rate": 0.0002210424306801593,
|
| 145 |
+
"loss": 3.5423,
|
| 146 |
+
"step": 11500
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.5,
|
| 150 |
+
"learning_rate": 0.0002200238019673475,
|
| 151 |
+
"loss": 3.5214,
|
| 152 |
+
"step": 12000
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.52,
|
| 156 |
+
"learning_rate": 0.00021900517325453573,
|
| 157 |
+
"loss": 3.4998,
|
| 158 |
+
"step": 12500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.54,
|
| 162 |
+
"learning_rate": 0.00021798654454172392,
|
| 163 |
+
"loss": 3.5231,
|
| 164 |
+
"step": 13000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.56,
|
| 168 |
+
"learning_rate": 0.00021696791582891212,
|
| 169 |
+
"loss": 3.4711,
|
| 170 |
+
"step": 13500
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.58,
|
| 174 |
+
"learning_rate": 0.00021594928711610032,
|
| 175 |
+
"loss": 3.5046,
|
| 176 |
+
"step": 14000
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.6,
|
| 180 |
+
"learning_rate": 0.00021493065840328854,
|
| 181 |
+
"loss": 3.5012,
|
| 182 |
+
"step": 14500
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.62,
|
| 186 |
+
"learning_rate": 0.00021391202969047674,
|
| 187 |
+
"loss": 3.5055,
|
| 188 |
+
"step": 15000
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.65,
|
| 192 |
+
"learning_rate": 0.00021289340097766494,
|
| 193 |
+
"loss": 3.4635,
|
| 194 |
+
"step": 15500
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.67,
|
| 198 |
+
"learning_rate": 0.00021187477226485317,
|
| 199 |
+
"loss": 3.5161,
|
| 200 |
+
"step": 16000
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.69,
|
| 204 |
+
"learning_rate": 0.00021085614355204136,
|
| 205 |
+
"loss": 3.4837,
|
| 206 |
+
"step": 16500
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.71,
|
| 210 |
+
"learning_rate": 0.00020983751483922956,
|
| 211 |
+
"loss": 3.4993,
|
| 212 |
+
"step": 17000
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.73,
|
| 216 |
+
"learning_rate": 0.00020881888612641776,
|
| 217 |
+
"loss": 3.4666,
|
| 218 |
+
"step": 17500
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.75,
|
| 222 |
+
"learning_rate": 0.00020780025741360598,
|
| 223 |
+
"loss": 3.4714,
|
| 224 |
+
"step": 18000
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.77,
|
| 228 |
+
"learning_rate": 0.00020678162870079418,
|
| 229 |
+
"loss": 3.4767,
|
| 230 |
+
"step": 18500
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.79,
|
| 234 |
+
"learning_rate": 0.00020576299998798238,
|
| 235 |
+
"loss": 3.4542,
|
| 236 |
+
"step": 19000
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.81,
|
| 240 |
+
"learning_rate": 0.0002047443712751706,
|
| 241 |
+
"loss": 3.4639,
|
| 242 |
+
"step": 19500
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.83,
|
| 246 |
+
"learning_rate": 0.0002037257425623588,
|
| 247 |
+
"loss": 3.4533,
|
| 248 |
+
"step": 20000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.85,
|
| 252 |
+
"learning_rate": 0.00020270711384954703,
|
| 253 |
+
"loss": 3.4635,
|
| 254 |
+
"step": 20500
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.88,
|
| 258 |
+
"learning_rate": 0.0002016884851367352,
|
| 259 |
+
"loss": 3.4334,
|
| 260 |
+
"step": 21000
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.9,
|
| 264 |
+
"learning_rate": 0.00020066985642392342,
|
| 265 |
+
"loss": 3.4604,
|
| 266 |
+
"step": 21500
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.92,
|
| 270 |
+
"learning_rate": 0.00019965122771111162,
|
| 271 |
+
"loss": 3.4712,
|
| 272 |
+
"step": 22000
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.94,
|
| 276 |
+
"learning_rate": 0.00019863259899829982,
|
| 277 |
+
"loss": 3.4693,
|
| 278 |
+
"step": 22500
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 0.96,
|
| 282 |
+
"learning_rate": 0.00019761397028548804,
|
| 283 |
+
"loss": 3.4672,
|
| 284 |
+
"step": 23000
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.98,
|
| 288 |
+
"learning_rate": 0.00019659534157267624,
|
| 289 |
+
"loss": 3.4405,
|
| 290 |
+
"step": 23500
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 1.0,
|
| 294 |
+
"learning_rate": 0.00019557671285986447,
|
| 295 |
+
"loss": 3.4546,
|
| 296 |
+
"step": 24000
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 1.0,
|
| 300 |
+
"eval_loss": 3.346637487411499,
|
| 301 |
+
"eval_runtime": 3451.9723,
|
| 302 |
+
"eval_samples_per_second": 0.724,
|
| 303 |
+
"step": 24000
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 1.02,
|
| 307 |
+
"learning_rate": 0.00019455808414705264,
|
| 308 |
+
"loss": 3.3935,
|
| 309 |
+
"step": 24500
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 1.04,
|
| 313 |
+
"learning_rate": 0.00019353945543424086,
|
| 314 |
+
"loss": 3.3975,
|
| 315 |
+
"step": 25000
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 1.06,
|
| 319 |
+
"learning_rate": 0.00019252082672142906,
|
| 320 |
+
"loss": 3.3951,
|
| 321 |
+
"step": 25500
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 1.08,
|
| 325 |
+
"learning_rate": 0.00019150219800861729,
|
| 326 |
+
"loss": 3.3671,
|
| 327 |
+
"step": 26000
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 1.1,
|
| 331 |
+
"learning_rate": 0.00019048356929580548,
|
| 332 |
+
"loss": 3.3708,
|
| 333 |
+
"step": 26500
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 1.12,
|
| 337 |
+
"learning_rate": 0.00018946494058299368,
|
| 338 |
+
"loss": 3.4103,
|
| 339 |
+
"step": 27000
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 1.15,
|
| 343 |
+
"learning_rate": 0.0001884463118701819,
|
| 344 |
+
"loss": 3.3958,
|
| 345 |
+
"step": 27500
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 1.17,
|
| 349 |
+
"learning_rate": 0.0001874276831573701,
|
| 350 |
+
"loss": 3.3796,
|
| 351 |
+
"step": 28000
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 1.19,
|
| 355 |
+
"learning_rate": 0.0001864090544445583,
|
| 356 |
+
"loss": 3.3851,
|
| 357 |
+
"step": 28500
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 1.21,
|
| 361 |
+
"learning_rate": 0.0001853904257317465,
|
| 362 |
+
"loss": 3.3876,
|
| 363 |
+
"step": 29000
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 1.23,
|
| 367 |
+
"learning_rate": 0.00018437179701893472,
|
| 368 |
+
"loss": 3.3663,
|
| 369 |
+
"step": 29500
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 1.25,
|
| 373 |
+
"learning_rate": 0.00018335316830612292,
|
| 374 |
+
"loss": 3.3685,
|
| 375 |
+
"step": 30000
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 1.27,
|
| 379 |
+
"learning_rate": 0.00018233453959331112,
|
| 380 |
+
"loss": 3.3707,
|
| 381 |
+
"step": 30500
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 1.29,
|
| 385 |
+
"learning_rate": 0.00018131591088049934,
|
| 386 |
+
"loss": 3.3691,
|
| 387 |
+
"step": 31000
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 1.31,
|
| 391 |
+
"learning_rate": 0.00018029728216768754,
|
| 392 |
+
"loss": 3.3854,
|
| 393 |
+
"step": 31500
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 1.33,
|
| 397 |
+
"learning_rate": 0.00017927865345487574,
|
| 398 |
+
"loss": 3.391,
|
| 399 |
+
"step": 32000
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 1.35,
|
| 403 |
+
"learning_rate": 0.00017826002474206394,
|
| 404 |
+
"loss": 3.3807,
|
| 405 |
+
"step": 32500
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 1.38,
|
| 409 |
+
"learning_rate": 0.00017724139602925216,
|
| 410 |
+
"loss": 3.3699,
|
| 411 |
+
"step": 33000
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 1.4,
|
| 415 |
+
"learning_rate": 0.00017622276731644036,
|
| 416 |
+
"loss": 3.3595,
|
| 417 |
+
"step": 33500
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 1.42,
|
| 421 |
+
"learning_rate": 0.00017520413860362856,
|
| 422 |
+
"loss": 3.3649,
|
| 423 |
+
"step": 34000
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 1.44,
|
| 427 |
+
"learning_rate": 0.00017418550989081678,
|
| 428 |
+
"loss": 3.362,
|
| 429 |
+
"step": 34500
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 1.46,
|
| 433 |
+
"learning_rate": 0.00017316688117800498,
|
| 434 |
+
"loss": 3.3782,
|
| 435 |
+
"step": 35000
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 1.48,
|
| 439 |
+
"learning_rate": 0.0001721482524651932,
|
| 440 |
+
"loss": 3.334,
|
| 441 |
+
"step": 35500
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 1.5,
|
| 445 |
+
"learning_rate": 0.00017112962375238138,
|
| 446 |
+
"loss": 3.3432,
|
| 447 |
+
"step": 36000
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 1.52,
|
| 451 |
+
"learning_rate": 0.0001701109950395696,
|
| 452 |
+
"loss": 3.3615,
|
| 453 |
+
"step": 36500
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 1.54,
|
| 457 |
+
"learning_rate": 0.0001690923663267578,
|
| 458 |
+
"loss": 3.3422,
|
| 459 |
+
"step": 37000
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.56,
|
| 463 |
+
"learning_rate": 0.00016807373761394603,
|
| 464 |
+
"loss": 3.3664,
|
| 465 |
+
"step": 37500
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 1.58,
|
| 469 |
+
"learning_rate": 0.00016705510890113422,
|
| 470 |
+
"loss": 3.3676,
|
| 471 |
+
"step": 38000
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 1.6,
|
| 475 |
+
"learning_rate": 0.00016603648018832242,
|
| 476 |
+
"loss": 3.3417,
|
| 477 |
+
"step": 38500
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 1.62,
|
| 481 |
+
"learning_rate": 0.00016501785147551065,
|
| 482 |
+
"loss": 3.3443,
|
| 483 |
+
"step": 39000
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 1.65,
|
| 487 |
+
"learning_rate": 0.00016399922276269882,
|
| 488 |
+
"loss": 3.3712,
|
| 489 |
+
"step": 39500
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 1.67,
|
| 493 |
+
"learning_rate": 0.00016298059404988704,
|
| 494 |
+
"loss": 3.3539,
|
| 495 |
+
"step": 40000
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 1.69,
|
| 499 |
+
"learning_rate": 0.00016196196533707524,
|
| 500 |
+
"loss": 3.3543,
|
| 501 |
+
"step": 40500
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 1.71,
|
| 505 |
+
"learning_rate": 0.00016094333662426346,
|
| 506 |
+
"loss": 3.3777,
|
| 507 |
+
"step": 41000
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 1.73,
|
| 511 |
+
"learning_rate": 0.00015992470791145166,
|
| 512 |
+
"loss": 3.3434,
|
| 513 |
+
"step": 41500
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 1.75,
|
| 517 |
+
"learning_rate": 0.00015890607919863986,
|
| 518 |
+
"loss": 3.3515,
|
| 519 |
+
"step": 42000
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 1.77,
|
| 523 |
+
"learning_rate": 0.00015788745048582809,
|
| 524 |
+
"loss": 3.3511,
|
| 525 |
+
"step": 42500
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 1.79,
|
| 529 |
+
"learning_rate": 0.00015686882177301628,
|
| 530 |
+
"loss": 3.3412,
|
| 531 |
+
"step": 43000
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 1.81,
|
| 535 |
+
"learning_rate": 0.00015585019306020448,
|
| 536 |
+
"loss": 3.3573,
|
| 537 |
+
"step": 43500
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 1.83,
|
| 541 |
+
"learning_rate": 0.00015483156434739268,
|
| 542 |
+
"loss": 3.3346,
|
| 543 |
+
"step": 44000
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 1.85,
|
| 547 |
+
"learning_rate": 0.0001538129356345809,
|
| 548 |
+
"loss": 3.3609,
|
| 549 |
+
"step": 44500
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.88,
|
| 553 |
+
"learning_rate": 0.0001527943069217691,
|
| 554 |
+
"loss": 3.3234,
|
| 555 |
+
"step": 45000
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 1.9,
|
| 559 |
+
"learning_rate": 0.0001517756782089573,
|
| 560 |
+
"loss": 3.3438,
|
| 561 |
+
"step": 45500
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 1.92,
|
| 565 |
+
"learning_rate": 0.00015075704949614552,
|
| 566 |
+
"loss": 3.3383,
|
| 567 |
+
"step": 46000
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.94,
|
| 571 |
+
"learning_rate": 0.00014973842078333372,
|
| 572 |
+
"loss": 3.3506,
|
| 573 |
+
"step": 46500
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 1.96,
|
| 577 |
+
"learning_rate": 0.00014871979207052192,
|
| 578 |
+
"loss": 3.3226,
|
| 579 |
+
"step": 47000
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 1.98,
|
| 583 |
+
"learning_rate": 0.00014770116335771012,
|
| 584 |
+
"loss": 3.3569,
|
| 585 |
+
"step": 47500
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 2.0,
|
| 589 |
+
"learning_rate": 0.00014668253464489834,
|
| 590 |
+
"loss": 3.3312,
|
| 591 |
+
"step": 48000
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 2.0,
|
| 595 |
+
"eval_loss": 3.2780921459198,
|
| 596 |
+
"eval_runtime": 3449.6819,
|
| 597 |
+
"eval_samples_per_second": 0.725,
|
| 598 |
+
"step": 48000
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 2.02,
|
| 602 |
+
"learning_rate": 0.00014566390593208654,
|
| 603 |
+
"loss": 3.2744,
|
| 604 |
+
"step": 48500
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 2.04,
|
| 608 |
+
"learning_rate": 0.00014464527721927474,
|
| 609 |
+
"loss": 3.27,
|
| 610 |
+
"step": 49000
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 2.06,
|
| 614 |
+
"learning_rate": 0.00014362664850646296,
|
| 615 |
+
"loss": 3.2869,
|
| 616 |
+
"step": 49500
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 2.08,
|
| 620 |
+
"learning_rate": 0.00014260801979365116,
|
| 621 |
+
"loss": 3.2813,
|
| 622 |
+
"step": 50000
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 2.1,
|
| 626 |
+
"learning_rate": 0.0001415893910808394,
|
| 627 |
+
"loss": 3.3014,
|
| 628 |
+
"step": 50500
|
| 629 |
+
},
|
| 630 |
+
{
|
| 631 |
+
"epoch": 2.12,
|
| 632 |
+
"learning_rate": 0.00014057076236802756,
|
| 633 |
+
"loss": 3.2948,
|
| 634 |
+
"step": 51000
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 2.15,
|
| 638 |
+
"learning_rate": 0.00013955213365521578,
|
| 639 |
+
"loss": 3.2821,
|
| 640 |
+
"step": 51500
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 2.17,
|
| 644 |
+
"learning_rate": 0.00013853350494240398,
|
| 645 |
+
"loss": 3.3228,
|
| 646 |
+
"step": 52000
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 2.19,
|
| 650 |
+
"learning_rate": 0.00013751487622959218,
|
| 651 |
+
"loss": 3.2893,
|
| 652 |
+
"step": 52500
|
| 653 |
+
},
|
| 654 |
+
{
|
| 655 |
+
"epoch": 2.21,
|
| 656 |
+
"learning_rate": 0.0001364962475167804,
|
| 657 |
+
"loss": 3.2784,
|
| 658 |
+
"step": 53000
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 2.23,
|
| 662 |
+
"learning_rate": 0.0001354776188039686,
|
| 663 |
+
"loss": 3.2613,
|
| 664 |
+
"step": 53500
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 2.25,
|
| 668 |
+
"learning_rate": 0.00013445899009115683,
|
| 669 |
+
"loss": 3.2574,
|
| 670 |
+
"step": 54000
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 2.27,
|
| 674 |
+
"learning_rate": 0.000133440361378345,
|
| 675 |
+
"loss": 3.2696,
|
| 676 |
+
"step": 54500
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 2.29,
|
| 680 |
+
"learning_rate": 0.00013242173266553322,
|
| 681 |
+
"loss": 3.2963,
|
| 682 |
+
"step": 55000
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 2.31,
|
| 686 |
+
"learning_rate": 0.00013140310395272142,
|
| 687 |
+
"loss": 3.2731,
|
| 688 |
+
"step": 55500
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 2.33,
|
| 692 |
+
"learning_rate": 0.00013038447523990964,
|
| 693 |
+
"loss": 3.2738,
|
| 694 |
+
"step": 56000
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 2.35,
|
| 698 |
+
"learning_rate": 0.00012936584652709784,
|
| 699 |
+
"loss": 3.2624,
|
| 700 |
+
"step": 56500
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 2.38,
|
| 704 |
+
"learning_rate": 0.00012834721781428604,
|
| 705 |
+
"loss": 3.2849,
|
| 706 |
+
"step": 57000
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 2.4,
|
| 710 |
+
"learning_rate": 0.00012732858910147427,
|
| 711 |
+
"loss": 3.2966,
|
| 712 |
+
"step": 57500
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 2.42,
|
| 716 |
+
"learning_rate": 0.00012630996038866246,
|
| 717 |
+
"loss": 3.2826,
|
| 718 |
+
"step": 58000
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 2.44,
|
| 722 |
+
"learning_rate": 0.00012529133167585066,
|
| 723 |
+
"loss": 3.2693,
|
| 724 |
+
"step": 58500
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 2.46,
|
| 728 |
+
"learning_rate": 0.00012427270296303886,
|
| 729 |
+
"loss": 3.3023,
|
| 730 |
+
"step": 59000
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 2.48,
|
| 734 |
+
"learning_rate": 0.00012325407425022708,
|
| 735 |
+
"loss": 3.2709,
|
| 736 |
+
"step": 59500
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"epoch": 2.5,
|
| 740 |
+
"learning_rate": 0.00012223544553741528,
|
| 741 |
+
"loss": 3.2935,
|
| 742 |
+
"step": 60000
|
| 743 |
+
},
|
| 744 |
+
{
|
| 745 |
+
"epoch": 2.52,
|
| 746 |
+
"learning_rate": 0.00012121681682460349,
|
| 747 |
+
"loss": 3.2927,
|
| 748 |
+
"step": 60500
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 2.54,
|
| 752 |
+
"learning_rate": 0.00012019818811179169,
|
| 753 |
+
"loss": 3.2921,
|
| 754 |
+
"step": 61000
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 2.56,
|
| 758 |
+
"learning_rate": 0.0001191795593989799,
|
| 759 |
+
"loss": 3.285,
|
| 760 |
+
"step": 61500
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 2.58,
|
| 764 |
+
"learning_rate": 0.0001181609306861681,
|
| 765 |
+
"loss": 3.2783,
|
| 766 |
+
"step": 62000
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 2.6,
|
| 770 |
+
"learning_rate": 0.00011714230197335631,
|
| 771 |
+
"loss": 3.2703,
|
| 772 |
+
"step": 62500
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 2.62,
|
| 776 |
+
"learning_rate": 0.00011612367326054451,
|
| 777 |
+
"loss": 3.2683,
|
| 778 |
+
"step": 63000
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 2.65,
|
| 782 |
+
"learning_rate": 0.00011510504454773272,
|
| 783 |
+
"loss": 3.2834,
|
| 784 |
+
"step": 63500
|
| 785 |
+
},
|
| 786 |
+
{
|
| 787 |
+
"epoch": 2.67,
|
| 788 |
+
"learning_rate": 0.00011408641583492093,
|
| 789 |
+
"loss": 3.2854,
|
| 790 |
+
"step": 64000
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 2.69,
|
| 794 |
+
"learning_rate": 0.00011306778712210914,
|
| 795 |
+
"loss": 3.2817,
|
| 796 |
+
"step": 64500
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 2.71,
|
| 800 |
+
"learning_rate": 0.00011204915840929734,
|
| 801 |
+
"loss": 3.2603,
|
| 802 |
+
"step": 65000
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 2.73,
|
| 806 |
+
"learning_rate": 0.00011103052969648555,
|
| 807 |
+
"loss": 3.2708,
|
| 808 |
+
"step": 65500
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 2.75,
|
| 812 |
+
"learning_rate": 0.00011001190098367375,
|
| 813 |
+
"loss": 3.265,
|
| 814 |
+
"step": 66000
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 2.77,
|
| 818 |
+
"learning_rate": 0.00010899327227086196,
|
| 819 |
+
"loss": 3.2474,
|
| 820 |
+
"step": 66500
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"epoch": 2.79,
|
| 824 |
+
"learning_rate": 0.00010797464355805016,
|
| 825 |
+
"loss": 3.2882,
|
| 826 |
+
"step": 67000
|
| 827 |
+
},
|
| 828 |
+
{
|
| 829 |
+
"epoch": 2.81,
|
| 830 |
+
"learning_rate": 0.00010695601484523837,
|
| 831 |
+
"loss": 3.274,
|
| 832 |
+
"step": 67500
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 2.83,
|
| 836 |
+
"learning_rate": 0.00010593738613242658,
|
| 837 |
+
"loss": 3.2589,
|
| 838 |
+
"step": 68000
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 2.85,
|
| 842 |
+
"learning_rate": 0.00010491875741961478,
|
| 843 |
+
"loss": 3.2984,
|
| 844 |
+
"step": 68500
|
| 845 |
+
},
|
| 846 |
+
{
|
| 847 |
+
"epoch": 2.88,
|
| 848 |
+
"learning_rate": 0.00010390012870680299,
|
| 849 |
+
"loss": 3.2629,
|
| 850 |
+
"step": 69000
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 2.9,
|
| 854 |
+
"learning_rate": 0.00010288149999399119,
|
| 855 |
+
"loss": 3.2833,
|
| 856 |
+
"step": 69500
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 2.92,
|
| 860 |
+
"learning_rate": 0.0001018628712811794,
|
| 861 |
+
"loss": 3.27,
|
| 862 |
+
"step": 70000
|
| 863 |
+
},
|
| 864 |
+
{
|
| 865 |
+
"epoch": 2.94,
|
| 866 |
+
"learning_rate": 0.0001008442425683676,
|
| 867 |
+
"loss": 3.2593,
|
| 868 |
+
"step": 70500
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 2.96,
|
| 872 |
+
"learning_rate": 9.982561385555581e-05,
|
| 873 |
+
"loss": 3.2753,
|
| 874 |
+
"step": 71000
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 2.98,
|
| 878 |
+
"learning_rate": 9.880698514274402e-05,
|
| 879 |
+
"loss": 3.2822,
|
| 880 |
+
"step": 71500
|
| 881 |
+
},
|
| 882 |
+
{
|
| 883 |
+
"epoch": 3.0,
|
| 884 |
+
"learning_rate": 9.778835642993223e-05,
|
| 885 |
+
"loss": 3.2595,
|
| 886 |
+
"step": 72000
|
| 887 |
+
},
|
| 888 |
+
{
|
| 889 |
+
"epoch": 3.0,
|
| 890 |
+
"eval_loss": 3.237163782119751,
|
| 891 |
+
"eval_runtime": 3454.1756,
|
| 892 |
+
"eval_samples_per_second": 0.724,
|
| 893 |
+
"step": 72000
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 3.02,
|
| 897 |
+
"learning_rate": 9.676972771712043e-05,
|
| 898 |
+
"loss": 3.2132,
|
| 899 |
+
"step": 72500
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 3.04,
|
| 903 |
+
"learning_rate": 9.575109900430864e-05,
|
| 904 |
+
"loss": 3.2369,
|
| 905 |
+
"step": 73000
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 3.06,
|
| 909 |
+
"learning_rate": 9.473247029149684e-05,
|
| 910 |
+
"loss": 3.1945,
|
| 911 |
+
"step": 73500
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 3.08,
|
| 915 |
+
"learning_rate": 9.371384157868505e-05,
|
| 916 |
+
"loss": 3.2169,
|
| 917 |
+
"step": 74000
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 3.1,
|
| 921 |
+
"learning_rate": 9.269521286587325e-05,
|
| 922 |
+
"loss": 3.2146,
|
| 923 |
+
"step": 74500
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 3.12,
|
| 927 |
+
"learning_rate": 9.167658415306146e-05,
|
| 928 |
+
"loss": 3.1977,
|
| 929 |
+
"step": 75000
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 3.15,
|
| 933 |
+
"learning_rate": 9.065795544024967e-05,
|
| 934 |
+
"loss": 3.2346,
|
| 935 |
+
"step": 75500
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 3.17,
|
| 939 |
+
"learning_rate": 8.963932672743787e-05,
|
| 940 |
+
"loss": 3.2214,
|
| 941 |
+
"step": 76000
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 3.19,
|
| 945 |
+
"learning_rate": 8.862069801462608e-05,
|
| 946 |
+
"loss": 3.2317,
|
| 947 |
+
"step": 76500
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 3.21,
|
| 951 |
+
"learning_rate": 8.760206930181428e-05,
|
| 952 |
+
"loss": 3.2254,
|
| 953 |
+
"step": 77000
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 3.23,
|
| 957 |
+
"learning_rate": 8.658344058900249e-05,
|
| 958 |
+
"loss": 3.1977,
|
| 959 |
+
"step": 77500
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 3.25,
|
| 963 |
+
"learning_rate": 8.556481187619069e-05,
|
| 964 |
+
"loss": 3.1996,
|
| 965 |
+
"step": 78000
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 3.27,
|
| 969 |
+
"learning_rate": 8.45461831633789e-05,
|
| 970 |
+
"loss": 3.2191,
|
| 971 |
+
"step": 78500
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 3.29,
|
| 975 |
+
"learning_rate": 8.352755445056711e-05,
|
| 976 |
+
"loss": 3.2199,
|
| 977 |
+
"step": 79000
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 3.31,
|
| 981 |
+
"learning_rate": 8.250892573775532e-05,
|
| 982 |
+
"loss": 3.2018,
|
| 983 |
+
"step": 79500
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 3.33,
|
| 987 |
+
"learning_rate": 8.149029702494352e-05,
|
| 988 |
+
"loss": 3.2097,
|
| 989 |
+
"step": 80000
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 3.35,
|
| 993 |
+
"learning_rate": 8.047166831213173e-05,
|
| 994 |
+
"loss": 3.2366,
|
| 995 |
+
"step": 80500
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 3.38,
|
| 999 |
+
"learning_rate": 7.945303959931993e-05,
|
| 1000 |
+
"loss": 3.2401,
|
| 1001 |
+
"step": 81000
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 3.4,
|
| 1005 |
+
"learning_rate": 7.843441088650814e-05,
|
| 1006 |
+
"loss": 3.2094,
|
| 1007 |
+
"step": 81500
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 3.42,
|
| 1011 |
+
"learning_rate": 7.741578217369634e-05,
|
| 1012 |
+
"loss": 3.2285,
|
| 1013 |
+
"step": 82000
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 3.44,
|
| 1017 |
+
"learning_rate": 7.639715346088455e-05,
|
| 1018 |
+
"loss": 3.2129,
|
| 1019 |
+
"step": 82500
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 3.46,
|
| 1023 |
+
"learning_rate": 7.537852474807276e-05,
|
| 1024 |
+
"loss": 3.2359,
|
| 1025 |
+
"step": 83000
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 3.48,
|
| 1029 |
+
"learning_rate": 7.435989603526096e-05,
|
| 1030 |
+
"loss": 3.2283,
|
| 1031 |
+
"step": 83500
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 3.5,
|
| 1035 |
+
"learning_rate": 7.334126732244917e-05,
|
| 1036 |
+
"loss": 3.201,
|
| 1037 |
+
"step": 84000
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 3.52,
|
| 1041 |
+
"learning_rate": 7.232263860963737e-05,
|
| 1042 |
+
"loss": 3.2187,
|
| 1043 |
+
"step": 84500
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 3.54,
|
| 1047 |
+
"learning_rate": 7.130400989682558e-05,
|
| 1048 |
+
"loss": 3.2368,
|
| 1049 |
+
"step": 85000
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 3.56,
|
| 1053 |
+
"learning_rate": 7.028538118401378e-05,
|
| 1054 |
+
"loss": 3.1903,
|
| 1055 |
+
"step": 85500
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 3.58,
|
| 1059 |
+
"learning_rate": 6.926675247120199e-05,
|
| 1060 |
+
"loss": 3.2153,
|
| 1061 |
+
"step": 86000
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 3.6,
|
| 1065 |
+
"learning_rate": 6.82481237583902e-05,
|
| 1066 |
+
"loss": 3.2102,
|
| 1067 |
+
"step": 86500
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 3.62,
|
| 1071 |
+
"learning_rate": 6.722949504557841e-05,
|
| 1072 |
+
"loss": 3.2342,
|
| 1073 |
+
"step": 87000
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 3.65,
|
| 1077 |
+
"learning_rate": 6.621086633276661e-05,
|
| 1078 |
+
"loss": 3.2238,
|
| 1079 |
+
"step": 87500
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 3.67,
|
| 1083 |
+
"learning_rate": 6.519223761995482e-05,
|
| 1084 |
+
"loss": 3.2205,
|
| 1085 |
+
"step": 88000
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 3.69,
|
| 1089 |
+
"learning_rate": 6.417360890714302e-05,
|
| 1090 |
+
"loss": 3.2328,
|
| 1091 |
+
"step": 88500
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 3.71,
|
| 1095 |
+
"learning_rate": 6.315498019433123e-05,
|
| 1096 |
+
"loss": 3.202,
|
| 1097 |
+
"step": 89000
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 3.73,
|
| 1101 |
+
"learning_rate": 6.213635148151943e-05,
|
| 1102 |
+
"loss": 3.2223,
|
| 1103 |
+
"step": 89500
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 3.75,
|
| 1107 |
+
"learning_rate": 6.111772276870764e-05,
|
| 1108 |
+
"loss": 3.1889,
|
| 1109 |
+
"step": 90000
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 3.77,
|
| 1113 |
+
"learning_rate": 6.0099094055895845e-05,
|
| 1114 |
+
"loss": 3.2399,
|
| 1115 |
+
"step": 90500
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 3.79,
|
| 1119 |
+
"learning_rate": 5.908046534308405e-05,
|
| 1120 |
+
"loss": 3.2285,
|
| 1121 |
+
"step": 91000
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 3.81,
|
| 1125 |
+
"learning_rate": 5.8061836630272255e-05,
|
| 1126 |
+
"loss": 3.2309,
|
| 1127 |
+
"step": 91500
|
| 1128 |
+
},
|
| 1129 |
+
{
|
| 1130 |
+
"epoch": 3.83,
|
| 1131 |
+
"learning_rate": 5.7043207917460466e-05,
|
| 1132 |
+
"loss": 3.2062,
|
| 1133 |
+
"step": 92000
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 3.85,
|
| 1137 |
+
"learning_rate": 5.602457920464867e-05,
|
| 1138 |
+
"loss": 3.2125,
|
| 1139 |
+
"step": 92500
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 3.88,
|
| 1143 |
+
"learning_rate": 5.5005950491836875e-05,
|
| 1144 |
+
"loss": 3.2107,
|
| 1145 |
+
"step": 93000
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 3.9,
|
| 1149 |
+
"learning_rate": 5.398732177902508e-05,
|
| 1150 |
+
"loss": 3.196,
|
| 1151 |
+
"step": 93500
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 3.92,
|
| 1155 |
+
"learning_rate": 5.296869306621329e-05,
|
| 1156 |
+
"loss": 3.2327,
|
| 1157 |
+
"step": 94000
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 3.94,
|
| 1161 |
+
"learning_rate": 5.1950064353401496e-05,
|
| 1162 |
+
"loss": 3.2208,
|
| 1163 |
+
"step": 94500
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 3.96,
|
| 1167 |
+
"learning_rate": 5.09314356405897e-05,
|
| 1168 |
+
"loss": 3.2465,
|
| 1169 |
+
"step": 95000
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 3.98,
|
| 1173 |
+
"learning_rate": 4.9912806927777905e-05,
|
| 1174 |
+
"loss": 3.2313,
|
| 1175 |
+
"step": 95500
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 4.0,
|
| 1179 |
+
"learning_rate": 4.889417821496612e-05,
|
| 1180 |
+
"loss": 3.2177,
|
| 1181 |
+
"step": 96000
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 4.0,
|
| 1185 |
+
"eval_loss": 3.212102174758911,
|
| 1186 |
+
"eval_runtime": 3461.5926,
|
| 1187 |
+
"eval_samples_per_second": 0.722,
|
| 1188 |
+
"step": 96000
|
| 1189 |
+
},
|
| 1190 |
+
{
|
| 1191 |
+
"epoch": 4.02,
|
| 1192 |
+
"learning_rate": 4.787554950215432e-05,
|
| 1193 |
+
"loss": 3.1662,
|
| 1194 |
+
"step": 96500
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 4.04,
|
| 1198 |
+
"learning_rate": 4.6856920789342526e-05,
|
| 1199 |
+
"loss": 3.1846,
|
| 1200 |
+
"step": 97000
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 4.06,
|
| 1204 |
+
"learning_rate": 4.583829207653073e-05,
|
| 1205 |
+
"loss": 3.1761,
|
| 1206 |
+
"step": 97500
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 4.08,
|
| 1210 |
+
"learning_rate": 4.4819663363718935e-05,
|
| 1211 |
+
"loss": 3.1539,
|
| 1212 |
+
"step": 98000
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"epoch": 4.1,
|
| 1216 |
+
"learning_rate": 4.380103465090714e-05,
|
| 1217 |
+
"loss": 3.1975,
|
| 1218 |
+
"step": 98500
|
| 1219 |
+
},
|
| 1220 |
+
{
|
| 1221 |
+
"epoch": 4.12,
|
| 1222 |
+
"learning_rate": 4.2782405938095344e-05,
|
| 1223 |
+
"loss": 3.1617,
|
| 1224 |
+
"step": 99000
|
| 1225 |
+
},
|
| 1226 |
+
{
|
| 1227 |
+
"epoch": 4.15,
|
| 1228 |
+
"learning_rate": 4.1763777225283556e-05,
|
| 1229 |
+
"loss": 3.1818,
|
| 1230 |
+
"step": 99500
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"epoch": 4.17,
|
| 1234 |
+
"learning_rate": 4.074514851247176e-05,
|
| 1235 |
+
"loss": 3.1817,
|
| 1236 |
+
"step": 100000
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"epoch": 4.19,
|
| 1240 |
+
"learning_rate": 3.9726519799659965e-05,
|
| 1241 |
+
"loss": 3.1782,
|
| 1242 |
+
"step": 100500
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 4.21,
|
| 1246 |
+
"learning_rate": 3.870789108684817e-05,
|
| 1247 |
+
"loss": 3.1821,
|
| 1248 |
+
"step": 101000
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 4.23,
|
| 1252 |
+
"learning_rate": 3.768926237403638e-05,
|
| 1253 |
+
"loss": 3.1572,
|
| 1254 |
+
"step": 101500
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 4.25,
|
| 1258 |
+
"learning_rate": 3.6670633661224586e-05,
|
| 1259 |
+
"loss": 3.1857,
|
| 1260 |
+
"step": 102000
|
| 1261 |
+
},
|
| 1262 |
+
{
|
| 1263 |
+
"epoch": 4.27,
|
| 1264 |
+
"learning_rate": 3.565200494841279e-05,
|
| 1265 |
+
"loss": 3.169,
|
| 1266 |
+
"step": 102500
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 4.29,
|
| 1270 |
+
"learning_rate": 3.4633376235600995e-05,
|
| 1271 |
+
"loss": 3.1594,
|
| 1272 |
+
"step": 103000
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 4.31,
|
| 1276 |
+
"learning_rate": 3.3614747522789206e-05,
|
| 1277 |
+
"loss": 3.1473,
|
| 1278 |
+
"step": 103500
|
| 1279 |
+
},
|
| 1280 |
+
{
|
| 1281 |
+
"epoch": 4.33,
|
| 1282 |
+
"learning_rate": 3.259611880997741e-05,
|
| 1283 |
+
"loss": 3.1631,
|
| 1284 |
+
"step": 104000
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 4.35,
|
| 1288 |
+
"learning_rate": 3.1577490097165616e-05,
|
| 1289 |
+
"loss": 3.1612,
|
| 1290 |
+
"step": 104500
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 4.38,
|
| 1294 |
+
"learning_rate": 3.055886138435382e-05,
|
| 1295 |
+
"loss": 3.1647,
|
| 1296 |
+
"step": 105000
|
| 1297 |
+
},
|
| 1298 |
+
{
|
| 1299 |
+
"epoch": 4.4,
|
| 1300 |
+
"learning_rate": 2.9540232671542025e-05,
|
| 1301 |
+
"loss": 3.136,
|
| 1302 |
+
"step": 105500
|
| 1303 |
+
},
|
| 1304 |
+
{
|
| 1305 |
+
"epoch": 4.42,
|
| 1306 |
+
"learning_rate": 2.8521603958730233e-05,
|
| 1307 |
+
"loss": 3.1624,
|
| 1308 |
+
"step": 106000
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 4.44,
|
| 1312 |
+
"learning_rate": 2.7502975245918438e-05,
|
| 1313 |
+
"loss": 3.1897,
|
| 1314 |
+
"step": 106500
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 4.46,
|
| 1318 |
+
"learning_rate": 2.6484346533106646e-05,
|
| 1319 |
+
"loss": 3.1641,
|
| 1320 |
+
"step": 107000
|
| 1321 |
+
},
|
| 1322 |
+
{
|
| 1323 |
+
"epoch": 4.48,
|
| 1324 |
+
"learning_rate": 2.546571782029485e-05,
|
| 1325 |
+
"loss": 3.1694,
|
| 1326 |
+
"step": 107500
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 4.5,
|
| 1330 |
+
"learning_rate": 2.444708910748306e-05,
|
| 1331 |
+
"loss": 3.1741,
|
| 1332 |
+
"step": 108000
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 4.52,
|
| 1336 |
+
"learning_rate": 2.3428460394671263e-05,
|
| 1337 |
+
"loss": 3.1793,
|
| 1338 |
+
"step": 108500
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 4.54,
|
| 1342 |
+
"learning_rate": 2.2409831681859468e-05,
|
| 1343 |
+
"loss": 3.1895,
|
| 1344 |
+
"step": 109000
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 4.56,
|
| 1348 |
+
"learning_rate": 2.1391202969047672e-05,
|
| 1349 |
+
"loss": 3.1417,
|
| 1350 |
+
"step": 109500
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 4.58,
|
| 1354 |
+
"learning_rate": 2.037257425623588e-05,
|
| 1355 |
+
"loss": 3.161,
|
| 1356 |
+
"step": 110000
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"epoch": 4.6,
|
| 1360 |
+
"learning_rate": 1.9353945543424085e-05,
|
| 1361 |
+
"loss": 3.2126,
|
| 1362 |
+
"step": 110500
|
| 1363 |
+
},
|
| 1364 |
+
{
|
| 1365 |
+
"epoch": 4.62,
|
| 1366 |
+
"learning_rate": 1.8335316830612293e-05,
|
| 1367 |
+
"loss": 3.172,
|
| 1368 |
+
"step": 111000
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 4.65,
|
| 1372 |
+
"learning_rate": 1.7316688117800498e-05,
|
| 1373 |
+
"loss": 3.1943,
|
| 1374 |
+
"step": 111500
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 4.67,
|
| 1378 |
+
"learning_rate": 1.6298059404988706e-05,
|
| 1379 |
+
"loss": 3.1911,
|
| 1380 |
+
"step": 112000
|
| 1381 |
+
},
|
| 1382 |
+
{
|
| 1383 |
+
"epoch": 4.69,
|
| 1384 |
+
"learning_rate": 1.527943069217691e-05,
|
| 1385 |
+
"loss": 3.1869,
|
| 1386 |
+
"step": 112500
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 4.71,
|
| 1390 |
+
"learning_rate": 1.4260801979365117e-05,
|
| 1391 |
+
"loss": 3.1707,
|
| 1392 |
+
"step": 113000
|
| 1393 |
+
},
|
| 1394 |
+
{
|
| 1395 |
+
"epoch": 4.73,
|
| 1396 |
+
"learning_rate": 1.3242173266553323e-05,
|
| 1397 |
+
"loss": 3.1826,
|
| 1398 |
+
"step": 113500
|
| 1399 |
+
},
|
| 1400 |
+
{
|
| 1401 |
+
"epoch": 4.75,
|
| 1402 |
+
"learning_rate": 1.222354455374153e-05,
|
| 1403 |
+
"loss": 3.1852,
|
| 1404 |
+
"step": 114000
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 4.77,
|
| 1408 |
+
"learning_rate": 1.1204915840929734e-05,
|
| 1409 |
+
"loss": 3.156,
|
| 1410 |
+
"step": 114500
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 4.79,
|
| 1414 |
+
"learning_rate": 1.018628712811794e-05,
|
| 1415 |
+
"loss": 3.1525,
|
| 1416 |
+
"step": 115000
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 4.81,
|
| 1420 |
+
"learning_rate": 9.167658415306146e-06,
|
| 1421 |
+
"loss": 3.1757,
|
| 1422 |
+
"step": 115500
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 4.83,
|
| 1426 |
+
"learning_rate": 8.149029702494353e-06,
|
| 1427 |
+
"loss": 3.1826,
|
| 1428 |
+
"step": 116000
|
| 1429 |
+
},
|
| 1430 |
+
{
|
| 1431 |
+
"epoch": 4.85,
|
| 1432 |
+
"learning_rate": 7.130400989682558e-06,
|
| 1433 |
+
"loss": 3.162,
|
| 1434 |
+
"step": 116500
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 4.88,
|
| 1438 |
+
"learning_rate": 6.111772276870765e-06,
|
| 1439 |
+
"loss": 3.1913,
|
| 1440 |
+
"step": 117000
|
| 1441 |
+
},
|
| 1442 |
+
{
|
| 1443 |
+
"epoch": 4.9,
|
| 1444 |
+
"learning_rate": 5.09314356405897e-06,
|
| 1445 |
+
"loss": 3.1639,
|
| 1446 |
+
"step": 117500
|
| 1447 |
+
},
|
| 1448 |
+
{
|
| 1449 |
+
"epoch": 4.92,
|
| 1450 |
+
"learning_rate": 4.074514851247176e-06,
|
| 1451 |
+
"loss": 3.1868,
|
| 1452 |
+
"step": 118000
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 4.94,
|
| 1456 |
+
"learning_rate": 3.0558861384353823e-06,
|
| 1457 |
+
"loss": 3.1968,
|
| 1458 |
+
"step": 118500
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 4.96,
|
| 1462 |
+
"learning_rate": 2.037257425623588e-06,
|
| 1463 |
+
"loss": 3.1899,
|
| 1464 |
+
"step": 119000
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 4.98,
|
| 1468 |
+
"learning_rate": 1.018628712811794e-06,
|
| 1469 |
+
"loss": 3.1581,
|
| 1470 |
+
"step": 119500
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"epoch": 5.0,
|
| 1474 |
+
"learning_rate": 0.0,
|
| 1475 |
+
"loss": 3.1818,
|
| 1476 |
+
"step": 120000
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 5.0,
|
| 1480 |
+
"eval_loss": 3.197413921356201,
|
| 1481 |
+
"eval_runtime": 3464.2347,
|
| 1482 |
+
"eval_samples_per_second": 0.722,
|
| 1483 |
+
"step": 120000
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 5.0,
|
| 1487 |
+
"step": 120000,
|
| 1488 |
+
"total_flos": 1.8161973854208e+18,
|
| 1489 |
+
"train_runtime": 74846.9771,
|
| 1490 |
+
"train_samples_per_second": 1.603
|
| 1491 |
+
}
|
| 1492 |
+
],
|
| 1493 |
+
"max_steps": 120000,
|
| 1494 |
+
"num_train_epochs": 5,
|
| 1495 |
+
"total_flos": 1.8161973854208e+18,
|
| 1496 |
+
"trial_name": null,
|
| 1497 |
+
"trial_params": null
|
| 1498 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3b5ca0bc3bede1cd59fbb153a215d19e4163fe6ea50214fd000f8812dfb75527
|
| 3 |
+
size 2351
|
vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|